Getting the subtype right: using bioinformatics to select the best ovarian cancer cell lines for research

Ovarian cancer is a devastating disease that ranks as the second most common and lethal gynecological cancer worldwide, causing over 200,000 global deaths annually. Unlike a single disease, ovarian cancer encompasses a group of cancers originating from the epithelial tissue, which is a thin lining covering the outer surface of an ovary. Epithelial ovarian cancer (EOC) consists of five major histological subtypes: high-grade serous, clear cell, endometrioid, mucinous, and low-grade serous ovarian carcinomas. These subtypes differ in their molecular features, clinical outcomes, and responses to chemotherapy. Therefore, understanding their pathophysiology and identifying therapeutic targets specific to each subtype is crucial.

One of the challenges in studying EOC is the lack of suitable in vitro models that accurately reflect the heterogeneity and characteristics of primary tumors. Cell lines are often used as in vitro models in cancer research due to their affordability and ease of manipulation. However, many studies utilizing EOC cell lines fail to consider the importance of subtype and the resemblance of cell lines to their corresponding primary tumors. This oversight can lead to inaccurate or misleading results that do not translate well to the clinical setting. Consequently, there is a need to evaluate the molecular features of EOC cell lines, compare them with primary tumor samples, stratify by subtype, and identify the most representative cell lines for each subtype most suitable for pre-clinical studies.

In a recent study published in Frontiers in Cell and Developmental Biology, Aideen McCabe, Oza Zaheed, Dr Simon McDade and Dr Kellie Dean from Queen’s University Belfast and University College Cork, addressed the suitability of cell lines as models for the major subtypes of EOC, the second most fatal gynecological malignancy. Their objective was to determine if the molecular features of EOC cell lines accurately reflect the gene expression profiles and molecular characteristics of their corresponding primary tumor subtypes.

The researchers employed a clustering method called non-negative matrix factorization (NMF) to analyze the gene expression profiles of 56 EOC cell lines. NMF effectively grouped the cell lines into five clusters, potentially corresponding to each of the five EOC subtypes. These clusters not only validated previous histological groupings but also classified previously unannotated cell lines. To confirm whether the cell lines exhibited the characteristic genomic alterations of each subtype, the researchers examined their mutational and copy number landscapes. For example, they found prevalent TP53 mutations in high-grade serous and low-grade serous cell lines, while ARID1A and PIK3CA mutations were common in clear cell and endometrioid cell lines, respectively. Additionally, they identified specific copy number alterations associated with each subtype.

Furthermore, the researchers compared the gene expression profiles of the cell lines with 93 primary tumor samples, stratified by subtype, to identify the cell lines most closely resembling each subtype. To provide researchers with panels of suitable subtype specific cell lines, they identified 16 cell lines that were most like high-grade serous ovarian cancer, 12 cell lines that closely resembled clear cell, and seven cell lines each for mucinous and low-grade serous ovarian cancer. The study also identified cell lines that displayed poor overall molecular similarity to EOC tumors, suggesting they should be avoided in pre-clinical studies. Importantly, cell line A2780, which is one of the most widely used in vitro models EOC according to the literature, was found to not be representative of any of the subtypes of EOC investigated.

While cell lines have their limitations, such as underrepresenting tumor heterogeneity and lacking the tumor microenvironment, they remain valuable models for cancer research due to their ease of use and affordability. Therefore, it is crucial to maximize the utility of these models by selecting cell lines that accurately represent the specific tumor subtypes under investigation.

In conclusion, despite advancements in technology and healthcare, ovarian cancer survival rates have seen only modest improvements in the past two decades. This poor prognosis can be attributed, in part, to the lack of consideration for EOC subtypes in pre-clinical studies. The study discussed here offers a potential reference for gene expression-based EOC studies, aiding researchers in selecting appropriate cell lines that represent EOC subtypes accurately. Furthermore, it emphasizes the necessity for additional datasets representing all subtypes to contribute to the expansion of subtype-specific research in the field of ovarian cancer. A deeper understanding of these diverse subtypes will undoubtedly lead to more targeted therapies, novel diagnostics, and improved survival rates for patients affected by EOC.

In a statement to Medicine Innovates Professor Kellie Dean said “During the course of Aideen’s work and based on discussions with collaborators, we realized that what Aideen found was valuable to the cancer research community – selecting the best cell lines to most closely represent the specific tumor subtype will lead to discoveries that take this into account”